Use of Uranium Decay Series
for Dating an Archaeological Smelting Site
MASsACHUSrs INSSTTE
OF TECHNOLOGY
by
FEB 0 8 2010
Violetta Wolf
LIBRARIES
Submitted to the Department of Materials
Science and Engineering in Partial
Fulfillment of the Requirements for the
Degree of
ARCHA/ES
Bachelor of Science
at the
Massachusetts Institute of Technology
June 2008
© 2008 Massachusetts Institute of Technology
All rights Reserved
Signature of A uthor .............................
Certified by.........
....... .. .............. ""
...
Department of Materials Science and Engineering
May 19, 2008
...........................................................................
Dorothy Hosler
Professor of Archaeology and Ancient Technology
Thesis Advisor
Accepted by .......................................................
....
............
Caroline A. Ross
Professor of Materials Science and Engineering
Chairman, Undergraduate Thesis Committee
Use of Uranium Decay Series
for Dating an Archaeological Smelting Site
by
Violetta Wolf
Submitted to the Department of Materials
Science and Engineering on May 19, 2008
in Partial Fulfillment of the Requirements
for the Degree of Bachelor of Science
in Archaeology and Materials
ABSTRACT
Through the identification of phases and their isotopic composition and variability, an
assessment of the applicability of uranium decay series dating to El Manchon slags was made.
El Manchon is the only Mesoamerican site to exhibit smelting technology. Uranium series
dating is typically used on geologically old natural material, but the El Manchon slags were not
suitable for other dating techniques. There are four requirements of uranium series dating:
measurable presence of appropriate isotopes, cogenetic phases within the material, isotopic
fractionation between phases, and the ability to physically separate the phases. This is the first
attempt to date archaeological material with the uranium series dating method.
Petrographic reflected light microscopy was used to identify the phases in the slags. Electron
beam microanalysis was used to identify the chemical composition of the identified phases. Ion
beam microanalysis was used to assess the isotopic fractionation between the phases. Electron
pulse disaggregation, hand-sorting, and magnetic separations were performed to separate the
phases.
The slags are composed of four different phases: a silica-melt phase, a quartz-like phase, a
copper phase, and a copper-iron-sulfide phase. These four phases are in abundant presence with
sufficient isotopic fractionation to make the El Manchon slags suitable for uranium series dating.
Thesis Advisor: Dorothy Hosler
Title: Professor of Archaeology and Ancient Technology
^Y _
Abstract
TABLE OF CONTENTS
3
Index of Tables
5
Index of Figures
6
Introduction
9
Problem Statement
9
ExtractiveMetallurgy: Smelting
11
West Mexican Metallurgy
15
El Manchon: A West Mexican Metal Smelting Site
17
Methods
21
The UraniumDecay Series
Secular Equilibrium
The Internal Isochron
Cogenetic Phases
21
22
23
25
Four Criteria
(1) Measurable Presence of Appropriate Isotopes
(2) Cogenetic Phases
(3) Isotopic Fractionation
(4) Physical Separation of Cogenetic Phases
29
29
30
30
30
Sectioning and Photomicrography
32
Electron Beam Microanalysis
34
Ion Beam Microanalysis
35
PhysicalSeparationof CogeneticPhases
36
Results
38
Initial Slag Selection
38
Sample Sectioning
39
Identification of Cogenetic Phases Through Photomicrography
40
CompositionalAnalysis of CogeneticPhases
44
Demonstrationof Fractionation
48
PhysicalSeparationof Cogenetic Phases
50
Conclusion
55
Archaeological Utility of Uranium Series Dating
55
Dating the El Manchon Slags
55
FurtherResearch
58
Appendix A: Mathematical Derivation of Isochron Slope
59
Appendix B: Initial Slag Sample Collection
61
Appendix C: Slag Sample Photomicrographs
67
Appendix D: Electron Beam Microprobe Data
74
Acknowledgements
108
References Cited
109
Index of Tables
Table 1.
Summary of initially selected slag samples.
32
Table 2.
Summary of sample sections.
41
Table 3.
Summary of data points taken by electron beam microanalysis.
46
Table 4.
Summary of data points gather be ion beam microanalysis.
50
Table 5.
Fractionation in Samples 24 and 25.
51
Table 6.
Summary of the physical separations of grains in Sample 24.
54
Table 7.
Summary of the slag samples used in this work.
61
Table 8.
Summary of data points analyzed by electron beam microprobe.
74
Index of Figures
Figure 1.
Simple bowl furnace of the type used in the Central
Andes, before and after smelting operation (courtesy of Prof. Heather
Lechtman).
Figure 2.
Rough slag Sample 15 from El Manchon.
Figure 3.
MIT 5258-A: Opaque section micrographs copper prill and matte (left)
and crystalline fayalite (right). (Sharp 2003).
Figure 4.
Map of location of El Manchon and topographic map of Mesoamerica,
showing the limits of the West Mexican Metalworking Zone
(Hosler 1994).
Figure 5.
A map of El Manchon from unpublished fieldwork by Prof. Dorothy
Hosler. Sectors 1 and 3 are habitation areas. Sector 2 is the smelting
area.
Figure 6.
Photograph of the slag pile at El Manchon courtesy of Prof. Dorothy
Hosler.
Figure 7.
A schematic of the uranium decay series, courtesy of Dr. Kenneth Sims.
Figure 8.
An example of a secular equilibrium condition plotted on isotope ratio
space.
Figure 9.
The isotope ratio space of a material that fractionated after melting and
subsequent solidification. The red data point represents the initial single
phase melt. Each blue data point represents a phase in the multiphase
solidified material, such as a solid metallurgical slags.
Figure 10.
The decay of cogenetic phases through four half-lives. The normalized
activities of the phases will continue to along a line with the passage of
time. This line, the isochron, changes slope from a slope of zero at the
time of formation, until the isotopes achieve secular equilibrium and lie
on the equiline.
27
Figure 11.
Example of an isochron derived from experimental data.
28
Figure 12.
The blue line represents the isochron with the theoretical unmixed
phases at either end. The red dot shows where a sample consisting of
half of each of the phases at either end of the blue isochron.
31
Figure 13.
A photomicrograph of sample 9B.
34
Figure 14.
An example of a smooth slag.
Figure 15.
An example of a rough slag.
Figure 16.
A photomicrograph of the silicon-melt phase, showing amorphous and
crystalline zones.
Figure 17.
A photomicrograph of dendritic growth in the silicon-melt phase.
Dendritic growth is characteristic of crystalline phases. This slag likely
cooled sufficiently for crystalline growth to begin, then solidified before
crystal growth was complete.
42
Figure 18.
A photomicrograph of the quartz-like phase. The silicon-melt phase is
present in the voids in the quartz-like phase.
43
Figure 19.
A photomicrograph showing a copper prill and surrounding copper-ironsulfide matte.
44
Figure 20.
45
A photomicrograph showing an atypically large example of a copperiron-sulfide matte. Copper and copper-iron-sulfide prills are also visible
in the silica-melt phase.
Figure 21.
The EDS spectrum for point 22.2.1. This is an example of a
characteristic spectrum for the silicon-melt phase.
47
Figure 22.
The EDS spectrum for point 23.2.2. This is an example of a
characteristic spectrum for the quartz-like phase.
47
Figure 23.
The EDS spectrum from point 25.2.1 with copper and iron peaks
identified. This is a characteristic spectrum for the copper-iron-sulfide
phase. Readings of silicon probably come from the silicon-melt phase
surrounding the prill.
48
Figure 24.
The EDS spectrum from point 25.2.1 with iron peaks identified. This is
a characteristic spectrum for the copper-iron-sulfide phase. Readings of
silicon probably come from the silicon-melt phase surrounding the prill.
48
Figure 25.
The EDS spectrum from point 25.1.2 with copper peaks identified. This
is a characteristic spectrum for the copper phase. Readings of silicon
probably come from the silicon-melt phase surrounding the prill.
49
Figure 26.
A portion of the purity separation of Sample 24.
52
Figure 27.
A portion of the impurities separation of Sample 24.
52
Figure 28.
A portion of the grains-with-inclusions separation of Sample 24.
Figure 29.
An example of theoretical data points in dark blue, with a constructed
isochron line in dark red. By comparing the slope of the isochron with
the slope of the equiline, a date can be determined for the sample.
Appendix B: Initial Slag Sample Collection
Figures 30 - 38.
Appendix C: Photomicrographs
Figures 39 - 50.
Appendix D: Electron Beam Microprobe Data
Figures 51 - 115.
Introduction
Problem Statement
My research goal is to determine if copper smelting slags from the archaeological site of El
Manchon, Guerrero, Mexico are suitable for uranium series dating through isotopic and
microstructural analysis. Uranium series dating is a radiometric dating method common to
geochemistry. This research is the first application of this technique to archaeological material,
specifically archaeological slags.
Uranium series dating of a material requires the presence in the material of uranium series
isotopes, the existence of cogenetic phases in the material, isotopic fractionation between the
cogenetic phases, and methods to separate the phases physically. These four criteria can be
assessed only by rigorous chemical and physical analysis through use of techniques from
materials science and geochemistry. Geologists use the uranium series dating method to date
carbonates and lavas, but these materials are completely dissimilar to slags. Accurate dates of
the El Manchon smelting slags will provide fundamental data by which to establish the
chronology of Mesoamerican extractive metallurgy. If such dates can be acquired through
uranium series dating, archaeology will be provided with a new method for dating slags,
worldwide.
El Manchon is the only known site in Mesoamerica with evidence of ancient copper smelting.
Professor Dorothy Hosler discovered the site in 1998 (Hosler 2003b, 2004) during her survey of
the Balsas River drainage to locate preHispanic copper smelting sites. While there is clear
archaeological evidence of metallurgical activities in preHispanic Mesoamerica (Hosler 1988a,
1988b, 1994, 2003a), the Spaniards and other Europeans also carried out smelting operations in
the New World after the invasion of Mexico in 1521 (Hosler, personal communication, April
2008). Because El Manchon is a unique example of a physically discrete copper smelting
9
facility within a larger archaeological site, it is essential that evidence of the smelting technology
be dated directly, rather than in association with other materials at the site.
Radiocarbon dates from the habitation areas at El Manchon agree with the chronology suggested
by pottery types recovered from the site, these dates range from 1350 CE to 1550 CE (Hosler
2003b, 2004). Wood and charcoal samples from test pits and horizontal excavations in the
habitation areas of the site were used to determine these carbon-14 dates. The pottery found in
the habitation area is of the Yestla Naranjo type, dating to the middle Postclassic period
(approximately 1300 CE). The organic material found in the smelting area has produced
ambiguous and unreliable accelerated mass spectrometry (AMS) carbon-14 dates between 750
CE and 1850 CE (Hosler 2003b, 2004).
The carbon-14 dates from the smelting area of El Manchon, on average, appear to be several
hundreds of years older than the carbon-14 dates from the habitation area (Hosler, personal
communication, April 2008). The dates may be inaccurate as a result of carbon-12
contamination from the molten ores during the smelting process (Sims, personal communication,
April 2008). While the molten ore was in contact with organic material within the furnace,
carbon could have been exchanged between the two materials. The ores would have had a much
lower carbon-14 to carbon-12 ratio than the organic material, as they are much older than the
organic material. If carbon were exchanged, the organic material would have an artificially low
carbon-14 to carbon-12 ratio. This could explain why the carbon-14 dates from the organic
material are older than the true age of the samples. This explanation would not explain the
determination of younger dates, however the smelting site was greatly disturbed by tree root
activity, therefore younger organic material could have been buried at the same depth as the
older smelting evidence even though it has no chronological correlation with the smelting
material. In general, the carbon-14 determinations made on organic material found at the
smelting site may be inaccurate because the organic material was not deposited along with the
slag.
Apart from carbon-14 dating, there are no other quantitative dating techniques appropriate for a
site of this age. Other radiometric dating techniques, such as potassium-argon dating, are
applicable only to much older material (Hedman 2007). Thermoluminescence dating and
electron spin resonance techniques are not yet calibrated to international standards. However, the
geological dating technique known as uranium series dating may be used to date slags directly.
Uranium series dating has never been applied to archaeological material, but it is well
understood from studies carried out on geological forms such as lava flows (Cooper et al 2003).
In the case of a slag, a uranium series date would correspond to when the slag solidified (Sims,
personal communication; Macfarlane, personal communication to Hosler 2002).
Extractive Metallurgy: Smelting
In both the ancient Old World and New World, smelting was the primary process by which metal
was extracted from ores. Both crucibles and furnaces served as the reaction vessels within which
a charge of ore and fuel was reduced to metal.
The process of smelting involves (1) the chemical separation of the metal from the non-metallic
components of the ore, and (2) the physical separation of the metal from the silicious gangue
(host rocks) associated geologically with the ore.
Most of the ore samples retrieved archaeologically at El Manchon and analyzed at CMRAE
(Sharp 2003) are oxides of copper, including malachite [Cu 2(OH) 2CO 3] and cuprite [Cu 2 0].
Metallic copper was extracted from these ores by direct reduction methods in which the furnaces
were charged with a mixture of comminuted ore and charcoal. Charcoal was identified
archaeologically at El Manchon, admixed with slags and with the remains of disassociated
furnaces (Hosler, personal communication, April 2008). The El Manchon furnaces were made of
stone, but none was sufficiently intact to determine the presence or absences of any refractory
lining material on the interior vessel walls (Hosler, personal communication).
A typical smelting direct reduction reaction for oxide ores of copper is given by the following
equation (the mineral formula does not represent stoichiometric components).
CuCO 3
+
COt
copper
carbon
carbonate
monoxide
heat
-
Cu
+
2CO2
copper
carbon
metal
dioxide
The carbon monoxide forms upon partial burning of the charcoal in the charge. Typically, the
smelted copper falls through the low-density, molten slag and collects at he furnace bottom
where it forms an ingot upon cooling.
Within the reaction vessel, smelting involves the physical separation of the smelted metal from
the unwanted gangue that forms as a silicious melt, known as slag. If the temperature of the
smelting operation is high enough to melt the slag so that it runs, the slag will float above the
much denser metal and can be removed easily (Figure 1).
blowpipe
furnace
-sme ted
:ning
a
Figure 1.Simple bowl furnace of the type used in the Central Andes, before and after smelting
operation (courtesy of Prof. Heather Lechtman).
Slags are important by-products of smelting. They function as scavengers, absorbing many of
the non-metallic impurities present in the ore, sloughed off the furnace walls, or contributed by
the charcoal ash (Bachmann 1982). Low-density, molten slags float above the smelted metal.
Thus they serve not only to purify the metal but to shield it from oxygen present in the furnace
environment so that is does not reoxidize.
Many of the slags accumulated in a large slag pile at El Manchon indicate that they were fully
molten by the end of the smelting process (Figure 2). Rachel Sharp's calculations of the melting
temperatures of the slags, based on their bulk chemical analyses, indicated a range that ran from
11500 to 1200 0 C (Sharp 2003). Her experimental determinations, based on reheating samples of
the slags, showed that they began to slump at 1150 0 C, although even at a temperature of 1250 0 C
many of the slags did not run freely. At El Manchon, strong prevailing winds may have
provided sufficient oxygen to bring the internal furnace temperature to the 1200 0 C range.
Figure 2. Rough slag Sample 15 from El Manchon.
Slags are oxides formed by the fusion of silica [SiO 2] with metallic oxides present in the ore that
is being smelted. Fayalite [2FeO-SiO 2] is one of the most common constituents of ancient
smelting slags. It is desirable, because it forms at a relatively low temperature (ca. 1178 0 C), the
eutectic temperature of the SiO2 -FeO (wistite) system. Electron beam microanalysis of the El
Manchon slags determined them to be fayalitic, composed of crystalline fayalite and a glassy
phase (Sharp 2003; Figure 3).
Figure 3. MIT 5258-A: Opaque section micrographs copper prill and matte (left) and crystalline
fayalite (right). (Sharp 2003).
West Mexican Metallurgy
Though virtually nothing is known about the technology of Mesoamerican copper smelting, the
metalworking traditions of the region are well documented. Metallurgy was introduced to West
Mexico from Andean South America at about 700 CE, after complex sociopolitical organization
had developed in Mexico (Hosler 1986, 1988a, 1988b, 1988c, 1990, 1994, 2003, 2005). By
dating the El Manchon slags accurately, this research can offer a firm date for the only smelting
technology identified in this complex technological history.
The metallurgy of West Mexico was not an indigenous development in there, but was introduced
from two separate regions of South America: the Andean area including Ecuador and Peru and
an area that included Colombia and lower Central America (Hosler 1986, 1988a, 1988b, 1988c,
1990, 1994, 2003, 2005). There is substantial evidence that contacts and technological
exchanges between northern Andean South America and West Mexico were occurring long
before metallurgy arrived in West Mexico. The oldest evidence derives from ceramic styles and
a unique style of tomb construction that dates to 200 BCE (Hosler 1994).
Andean and Colombian metallurgies were most likely transferred to West Mexico through
technological exchange by means of a maritime route (Hosler 1986, 1994). Evidence of the
South American style of metalworking found in West Mexico is not present in any land area
between the two localities. In addition, the balsa log rafts that would have been used to conduct
maritime trade have been proven capable of such a voyage (Dewan 2008). While the West
Mexican metallurgical tradition clearly originated in South America, West Mexican metallurgy
did not replicate the metallurgies of the Andean zone, but integrated the two different
technologies into a unique metallurgical tradition. Hosler has divided this tradition into two
periods (Hosler 1986, 1988a, 1988b, 1988c, 1990, 1994, 2003, 2005). The first ranges from 600
CE to 1200/1300 CE, and the second period ranges from 1200/1300 CE to the Spanish invasion
in 1521 CE.
During the first period of West Mexican metalworking, few tools were produced. Metal was
used primarily to make elite objects (Hosler 1994). Smiths worked in gold, silver, copper, and a
copper-arsenic alloy to make fishhooks, awls, bells with wirework decoration, needles, rings, and
tweezers (Hosler 1994). Copper was the most common metal used for these artifacts (Hosler,
personal communication, March 2008). Bells were made by lost wax casting in the Colombian
tradition, while all the other objects were produced by cold-working metal blanks in the Andean
tradition (Hosler 1994). Smiths focused on the production of bells, as the sound-making
properties of metal were unique in the suite of materials technologies practiced by the
Mesoamericans (Hosler 1994).
The second period of West Mexican metallurgy is marked by a radical increase in the number of
alloys employed, which allowed for the redesign of many artifacts (Hosler 1986, 1988a, 1988b,
1988c, 1990, 1994, 2003, 2005). These new designs took full advantage of the material
properties of the alloys and pushed them to their mechanical limits. Tweezers began to appear
with a stylistic shell shape design (Hosler 1994). Several new artifact types also appeared: axes,
axe-monies, and sheet metal ornaments (Hosler 1994). Metalworkers used copper-arsenic-silver,
copper-arsenic-tin, copper-gold, copper-silver, copper-tin, copper-silver-gold, and a wider range
of copper-arsenic alloys (Hosler 1994). These new alloys are notable as they allowed for
refinement of design, and gave smiths control over the colors of the artifacts they produced
(Hosler 1994). The ability to manipulate the colors of metal artifacts was an important
development in the ritual use of metal, as the color engineering of these artifacts was purely an
aesthetic choice. In addition to using previously developed techniques, West Mexican smiths
began to hot work their products (Hosler 1986, 1988a, 1988b, 1988c, 1990, 1994, 2003, 2005).
El Manchon: A West Mexican Metal Smelting Site
The most distinguishing feature of el Manchon is that it is the only smelting site thus far located,
mapped, and partially excavated in Mesoamerica, a region where extensive metalworking
complexity has been identified and studied (Hosler 1986, 1988a, 1988b, 1988c, 1990, 1994,
2003, 2005). The site lies within the West Mexican Metalworking Zone, a zone that has a high
density of copper ore deposits (Figure 4).
J
's
,
\(mmI\
'I
;9 ,
4 t
~O4
CENTRAL
MEXICAN
PLATEAU
GULF Of MEXICO
TAULA
SULA
PACIFIC OCEAN
Transverse Vocanc Axis
'
' \;i~$~
/
0
10
200
3_0
I
40km
i
.
ri
.
El Manchon
Figure 4. Map of location of El Manchon and topographic map of Mesoamerica, showing the
limits of the West Mexican Metalworking Zone (Hosler 1994).
Because the site shows clear evidence of an area devoted to copper smelting, archaeologists
believe the primary activity at El Manchon was exploitation of metal resources (Hosler, personal
communication, March 2008). Though the site is located near many copper ore deposits, the
location of the deposit that was mined has not yet been identified. El Manchon must have
depended on outside resources for its subsistence base, as the area surrounding the site has a
shallow level of eroded soil that would not have supported agriculture (Hosler, personal
communication, March 2008).
Professor Dorothy Hosler of the Massachusetts Institute of Technology located and began to
excavate the site of El Manchon in 1998, during her survey of the municipio of Coyuca de
Catalan, Guerrero (Hosler, 1999, 2001-2003). El Manchon is located on a small tributary of the
Balsas River at an elevation of 1400 m above sea level. The site itself consists of one large
smelting area with two adjacent living areas (Figure 5).
F
E
C
ACUMULACUlN DE E5AI)RIA
u
r.
i
k~!Ll~-:
""~'"
g
-r'i?--1
il*C~h..."~
5ITIO LAFUNDICICNDE EMANCHON CUERRERC
Figure 5. A map of El Manchon from unpublished fieldwork by Prof. Dorothy Hosler. Sectors 1
and 3 are habitation areas. Sector 2 is the smelting area.
The smelting sector measures 40 X 50 m2 (Hosler, personal communication April 2008). Seven
smelting furnaces have been excavated within the smelting area, but the slag samples studied in
this thesis were found in another feature. They were collected from a slag pile that was eroding
down a ravine (Figure 6).
".
%4
EJTi
a
::
~
x-9
Figure 6.Photograph of the slag pile at El Manchon courtesy of Prof. Dorothy Hosler.
An undergraduate DMSE-MIT thesis by Rachel Sharp in 2001 was the first work undertaken to
analyze el Manchon slags and ores, prior to this thesis. Sharp's initial characterization of the
slags was the starting point of this investigation. In her work, she carried out bulk analyses of
the slags and developed a morphological classification system for the slag. She divided the slags
into two main categories: smooth and rough, based on their surface morphologies (Sharp 2001).
The most relevant result of Sharp's work for this thesis was that the slags were self-fluxing
(Sharp 2001), indicating that no extraneous materials had been added to the ores to facilitate
smelting. This allowed me to assume that the smelting furnaces were closed systems.
Currently many archaeologists use the designations BCE (before the common era), to replace
B.C., and CE (the common era), to replace A.D.
Methods
The Uranium Decay Series
238
U decays to
2
06Pb
through a series of alpha and beta decays with a half-life of 4.47 billion
years (Figure 7), (Ivanovich and Harmon 1982; Faure and Mensing 2005). This overarching
decay chain consists of several radioactive daughter isotopes with highly variable half-lives.
These daughter isotopes and the parent isotope 238U exist as trace elements in almost all
materials. The isotopes are usually present only as minor impurities and do not affect the
structure or chemical properties of a material significantly.
238-U Decay Series
Secular Equilibrium:
An = A22 6Ra
a
A= NX
Pa 2
Th
Gas
#ti/
s
ky
241 ds
2
L!j
Rn
At
Si21%4
1.*
Po
Po0
210
PKey
.10 "in
.,
S1.,
218
A = activity
,.
210
Key
N = abundance
..-
X= decay constant
i.-
tl/2= half-life
-IwI
Figure 7. A schematic of the uranium decay series, courtesy of Dr. Kenneth Sims
21
2 34
U, 23 0Th, and
226
Ra are the longest-lived daughter isotopes of 238 U with half-lives on the order
of thousands of years. The shortest lived isotopes, such as 218Rn and 2 14 po, have half-lives of a
few microseconds. These quickly decaying isotopes are not practical for experimental
measurements, as analytical error quickly outweighs the abundances of the isotopes. For
experimental dating,
2 34
U,
230 Th,
the case of the El Manchon slags,
and
226
Ra are used because of their relatively long half-lives. In
23 0
Th and 22 6Ra are used for dating. (For more information
regarding isotopic choice, please refer to the chapter on Methods, section on Four Criteria).
Secular Equilibrium
Uranium series dating is based on the fact that the decay chain reaches an equilibrium state of
decay after sufficient time has passed (Ivanovich and Harmon 1982; Faure and Mensing 2005).
Before the decay reaches equilibrium it is in a state of disequilibrium, the degree of which can be
measured to determine the age of the material being dated.
This type of equilibrium is known as secular equilibrium, in which the activity of the daughter
isotope is equal to that of the parent isotope (Equation 1). The activity (Equation 2) of an isotope
is equivalent to its abundance (how much of the isotope is present) multiplied by its decay
constant (Equation 3) (the rate at which the radioactive isotope is decaying) (Figure 7).
(1)
A230Th = A226Ra
(2)
A = NX
(3)
=ln(2)
tl/2
A = activity
N = abundance
k = decay constant
tl/2 = half-life
Because
238U
has such a long half-life, the abundance of 238U is considered infinite even though
no new 238U is being created. The half-life of 238U is 4.47 billion years, while
226Ra
has a half-
life of 1,602 years and 230Th has a half-life of 7,520 years (Ivanovich and Harmon 1982; Faure
22
and Mensing 2005). It takes about six half-lives for an isotope to achieve a low enough degree
of disequilibrium with its daughter isotope so as to be measurably in secular equilibrium (Sims,
personal communication, April 2008). Thus, the simple presence or absence of disequilibria can
put limits on the age of a material. For example, if the material being analyzed has a measurable
excess of 230Th relative to its parent
238 U,
then the age of the material is less than 450,000 years.
Similarly, if the sample has a significant excess (or deficit) of 226Ra relative to
230Th,
then its age
is less than 8,000 years.
The Internal Isochron
Tighter constraints can be placed on the age of a material by using an internal isochron method.
The isotopes
23 0Th
and 226Ra provide an example. If the normalized activities of 23 0Th and 226Ra
are plotted against one another, the plot results in a straight line known as the equiline (Figure 8).
Equiline Example
3
-
..................
2.5
-
Equiline
S. Material older than
8,000 years
1.5
* 2,500 year old slag
* 7,000 year old bone
0.5
0
0.5
1
1.5
2
2.5
3
((230Th)/BaE-3) {(dpm/g)/(ug/g)}
Figure 8. An example of a secular equilibrium condition plotted on isotope ratio space.
23
The activities of the two isotopes are normalized with the stable isotope
13 8
Ba, which, as a
geochemical proxy for radium, allows one to account fro any radium in a material that is not
associated with the decay of 230Th (Cooper et al 2003). The normalized ratios of these two
isotopes within a material will lie along the equiline if they have achieved secular equilibrium. If
the isotope ratios of a material 8,000 years or older were plotted as a single point on Figure 8, the
point would lie somewhere on the blue equiline because it takes 8,000 years for
226
Ra and 230 Th
to achieve secular equilibrium. If the isotope ratios of a material younger than 8,000 years were
plotted on Figure 8, the point could lie anywhere but on the equiline. For example, the blue
point in Figure 8 could represent a piece of 7,000-year-old hominid bone, and the maroon point
could represent a sample of 2,500-year-old metallurgical slag.
As time passes and the isotopes continue to decay, the data points approach the equiline by
moving in a predictable way in the y-direction. Because
226Ra,
2 30
Th has a much longer half-life than
the abundance of 230 Th is considered to remain constant during the time it takes 226Ra to
achieve secular equilibrium. Therefore, as time passes, the activity of 226Ra (which is plotted on
the y-axis) changes, but the activity of 230Th (which is plotted on the x-axis) does not change.
Thus, in uranium series dating, the y-axis is a proxy for time.
The direction that data points move is dependent on whether the material is rich or poor in the
226Ra.
If the material is
isotope,
230
22 6Ra-rich,
this means that there is more 226Ra compared to its parent
Th, than if the material were to be in secular equilibrium. If the material is
226
Ra-
poor, this means that there is less 226Ra compared to 230Th than if the material were to be in
secular equilibrium. Thus, in Figure 8, the dark blue data point would move up towards the blue
equiline and the maroon data point would move down towards the blue equiline. Despite the
predictability of the decay of isotopes in disequilibria, it is impossible to date a single data point
without knowing the isotopic composition of the material at the time of its formation. In
geological materials, predicting the isotopic ratio at the time of formation is difficult. In
archaeological materials, including the El Manchon slags, it is not possible. In geological
situations, one can make assumptions about the isotopic composition at the time of formation
based on geochemical understanding of the origin of the material. This is clearly not the case for
metallurgically-produced slags.
Cogenetic Phases
Although a single phase cannot be dated using uranium decay series, a set of cogenetic
phases can be dated. When a material is in a liquid or gas phase, the uranium decay series
isotopes are free to mix evenly throughout the material. In such a case, measurement of the
isotopic composition of the single phase material would yield only one data point plotted on
((226Ra)/Ba vs. (230Th)/Ba) space. When the material solidifies, however, if several solid phases
form the isotopes will fractionate into these phases because of the different chemical behaviors
of both the isotopes and the solid phases (Ivanovich and Harmon 1982). For example, uranium
and thorium both have larger ionic radii than radium, which impedes them from being included
in crystalline structures (Ivanovich and Harmon 1982). Uranium and thorium move
preferentially into amorphous phases because of their large radii, which results in fractionation of
these isotopes (Figure 9).
Fractionation
-3
2.5
-
2
Equiline
-Initial melt
1.5
*-Material after
solidification
0.5
0
0.5
1
1.5
2
2.5
3
((230Th)/BaE-3) {(dpm/g)/(ug/g)}
Figure 9. The isotope ratio space of a material that fractionated after melting and subsequent
solidification. The red data point represents the initial single phase melt. Each blue data point
represents a phase in the multiphase solidified material, such as a solid metallurgical slags.
Immediately after fractionation, the activities of the isotopes in the fractionated phases must
always add up to the initial activities of the melt. Since no new isotopes are formed when a
material melts, the existing isotopes simply mix freely. Fractionation results in phases all of
which are the same age but that are not all in secular equilibrium. Because they are constrained
to the same age (the y-direction in Figure 9), the fractionated phases all lie on a horizontal line
that intercepts the point on the equiline for the initial melt (the red point in Figure 9). As the
isotopes decay through time, each isotopic ratio will move parallel to the y-axis towards the blue
equiline. This behavior is the same as that described in Figure 8. In Figure 9 however,
individual values of different phases in the same material are changing, as distinct from the case
shown in Figure 8, where the bulk values for two different materials are changing.
M___
M__
M-
As noted above, the
226
Ra/Ba ratios of phases that are
lower values of this ratio, whereas the
226
22 6
Ra-rich will always change towards
Ra/Ba ratios of phases that are
226Ra-poor
will always
change towards higher values of the ratio. Like all radioactive decay, the decay of 226Ra is
exponential. After one half-life has passed, half of the initial abundance of the isotope remains.
By plotting the normalized activities of isotopes in disequilibrium through several half-lives, it
becomes clear that the data points for a multiphase material will always lie on a line that passes
through the initial melt point (Figure 10).
Isochron: Half-Life Decay
4"-
Fractionation
S2.5
-
Equiline
Time zero
-
After one half-life
After two half-lives
After three half-lives
,a1.5
1
0.5
0
0
1
2
3
((230Th)/BaE-3) {(dpm/g)/(ug/g)}
Figure 10. The decay of cogenetic phases through four half-lives. The normalized activities
of the phases will continue to along a line with the passage of time. This line, the isochron,
changes slope from a slope of zero at the time of formation, until the isotopes achieve
secular equilibrium and lie on the equiline.
The line defined by the isotopic compositions of a set of cogenetic phases at a single time is
known as an isochron (which means "equal time" or "same time"). In radiometric analysis, the
isochron is a line of best fit due to analytical error and deviation from ideal behavior (Figure 11).
Because the decay rates of 22 6Ra and 230Th are known, the slope of the isochron changes in a
predictable way through time. This allows the age of a sample to be calculated from the slope of
the isochron. The specific relationship is
m = 1 - e"X
(4)
m = slope of the isochron
X = decay constant of 226Ra
t = time
For a full derivation of this equation, please refer to appendix A.
Experimentally Derived Isochron
3
0%
32.5
2
1.5
!o
-
Equiline
-
Experimentally derived
isochron
.
0.5
1
2
((230Th)/BaE-3) {(dpm/g)/(ug/g)}
Figure 11. Example of an isochron derived from experimental data.
Four Criteria
The metallurgical slags must meet four criteria to be eligible for dating by the uranium decay
series method: (1) measurable presence of appropriate isotopes; (2) cogenetic phases within the
slag; (3) isotopic fractionation between phases; (4) the physical separability of the phases. Work
done by Christian Miller, a graduate student at WHOI, under the supervision of Dr. Kenneth
Sims of the Department of Geology and Geophysics at WHOI, determined that the El Manchon
slags contain measurable presences of the appropriate isotopes. I used petrographic reflected
light microscopy and electron beam microanalysis to assess the presence of phases within the
slags. Using an ion beam microprobe, I assessed the isotopic variation between the phases. By
separating the phases experimentally through electron pulse disaggregation, hand sorting, and
magnetic separations, I determined the relative ease of phase separations.
(1) Measurable Presence of Appropriate Isotopes
While uranium and its daughter elements are present as traces in most materials, the isotopes
used in this study had to be present in quantities sufficient to be measured by mass spectrometry.
Christian Miller demonstrated that there are sufficient abundances of the uranium series isotopes
in the El Manchon slags to permit measurements, but his most important conclusion was
determining which parent and daughter isotope should be used in this research.
The isotope space used to produce an isochron is determined by estimating the age of the
material. From the archaeological evidence at El Manchon, the slags are estimated to be
approximately 500 to 700 years old. Miller used quantitative isotope counts to confirm this age
range. He demonstrated that
210Pb
was in secular equilibrium and that 226Ra was not.
210Pb
has a
half-life of 22 years and takes about 100 years to achieve secular equilibrium. Miller compared
the activity of 210Pb to the activity of its parent
uranium decay chain).
210Pb
226Ra
226Ra
(refer to Figure 7 for a schematic of the
is not the direct parent of 2 10Pb, but it is the youngest parent of
with a half-life longer than three days. Because the activities of 2 10Pb and 226Ra were
equal, the slags had to be older than 100 years. Miller also compared the activity of 226Ra with
the activity of its parent, 230Th.
226Ra
has a half-life of 1600 years and takes approximately 8,000
years to achieve secular equilibrium. Miller's work showed that the activities of 22 6Ra and 230Th
were unequal, therefore the two isotopes were not in secular equilibrium. Because
23 0
226Ra
and
Th are not in secular equilibrium, they can be used to construct an isochron for the El
Manchon slags.
(2) Cogenetic Phases
The second criterion for dating the slags by the uranium decay method is that they contain
cogenetic phases. Cogenetic phases are phases that share the same parent material and time of
formation. The parent material for the El Manchon slags was the smelted ore of which the slag
is a byproduct. If the slags contain only one phase, yielding a single set of isotopic ratios, no
isochron can by plotted. To generate a multi-point isochron, the slag sample must have multiple
phases. Each phase would determine a point on the isochron. The more data points present, the
better the linear fit and error calculation between the experimental and ideal isochrons.
(3) Isotopic Fractionation
Even though cogenetic phases produce a more accurate isochron, they must also fulfill the third
criterion, that they exhibit isotopic fractionation. Even if the cogenetic phases are present, if the
phases do not exhibit fractionation, the isochron will still consist of only a single point. As
discussed in the uranium series background section, since uranium isotopes have different
chemical properties they tend to fractionate among the phases within a multiphase material. A
more probable issue would arise if some of the phases exhibited almost no isotopic presence.
For example, some materials, such as quartz, have internal structures that accommodate very
little of the uranium series isotopes (Sims, personal communication, March 2008).
(4) Physical Separation of Cogenetic Phases
The final requirement for dating the El Manchon slags by the uranium series method is the
physical separability of the cogenetic phases. In order to measure the abundances of isotopes
30
within different phases, the phases must be physically separate from one another. Ideally, the
phases would be separated perfectly. In reality, the separation of phases is rarely complete. This
will not affect the age derived from the resulting isochron, however, as the mixed phases will lie
on the same isochron as the one generated by fully separated phases (Figure 12).
Isochron Generated by Mixed Phases
-
1.5
Isochron
* Non mixed phases
* Mixed phase
0.5
0
0
1
2
3
((230Th)/BaE-3) {(dpm/g)/(ug/g)}
Figure 12. The blue line represents the isochron with the theoretical unmixed phases at either end.
The red dot shows where a sample consisting of half of each of the phases at either end of the
blue isochron.
The mixing of phases will not affect the age calculated from the isochron, but individual data
points that represent the mixed phases will lie closer together than if the phases had been
unmixed. Referring to Figure 12, it is clear that the two blue end points, each of which
represents a pure phase, are the farthest apart they can be along the isochron line. Any mixtures
of the two pure phases will locate closer together along the isochron than the two pure phases.
Phase separations that consist of multiple phases will yield the same age as will pure phases but
the result is an isochron that consists of closely clustered data points, making the linear fit more
uncertain.
Sectioning and Photomicrography
Table 1. Summary of initially selected slag samples.
Site Location
Photo-
Number
El Manchon 2001
micrographed
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Sector 2 (smelting area)
Quadrant S2, El
Slag Pit Level 3, Bag 1
Slag Pit Level 3, Bag 1
Slag Pit Level 3, Bag 1
Slag Pit Level 3, Bag 1
Slag Pit Level 3, Bag 1
Slag Pit Level 3, Bag 2
Slag Pit Level 3, Bag 2
Slag Pit Level 3, Bag 2
Slag Pit Level 3, Bag 2
Slag Pit Level 3, Bag 2
Slag Pit Level 3, Bag 2
Slag Pit Level 4, Bag 3
Slag Pit Level 4, Bag 3
Slag Pit Level 4, Bag 3
Slag Pit Level 4, Bag 3
Slag Pit Level 4, Bag 3
Slag Pit Level 4, Bag 3
Slag Pit Level 4, Bag 3
Slag Pit Level 4, Bag 3
Slag Pit Level 4, Bag 3
Slag Pit Level 3, Bag 2
Slag Pit Level 3, Bag 2
Slag Pit Level 3, Bag 2
Slag Pit Level 3, Bag 2
Slag Pit Level 3, Bag 2
Sample
Type
Smooth
Rough
Rough
Smooth
Smooth
Smooth
Smooth
Smooth
Rough
Rough
Rough
Smooth
Smooth
Rough
Rough
Rough
Smooth
Rough
Smooth
Rough
Smooth
Rough
Rough
Rough
Rough
X
X
X
X
X
X
An initial collection of 25 different slags were selected from the material excavated from the slag
pile in the smelting sector ofEl Manchon (Figure 6). Slags that would be classified as rough,
smooth, or were not easily classifiable under the Sharp classification were selected (Sharp 2003).
Overall, the sample collection was chosen to be representative of the slags excavated from El
Manchon (refer to Appendix B). These slags were washed, sketched, and photographed with a
CoolPix 5400 digital camera set to macro mode, with 5M pixel proportions and high quality.
The photographs had a background of ground glass over white paper. In order to investigate the
internal morphology of Samples 4, 6, 9, 10, 14, and 15 sections were prepared for petrographic
reflected light microscopy. Samples 9, 10, and 15 were large enough to be sectioned twice.
The samples were sectioned with a diamond saw in the Ceramics Facility of the Center for
Materials Research in Archaeology and Ethnography at MIT. The samples were cut to fit into 1"
diameter round aluminum rings. They were mounted in the aluminum rings with epoxy. Using
an automatic sample polisher, the samples were polished with 240 grit silicon carbide until the
sample and epoxy were flush with the sample mount. The samples were then polished with 320
grit silicon carbide for five minutes, 400 grit silicon carbide for five minutes, 5 micron alumina
for thirty minutes, 1 micron alumina for thirty minutes, and finally 0.02 micron colloidal silica
for thirty minutes. This polishing created a smooth surface with sufficient relief to make the
internal microstructure clearly visible with a microscope. Photomicrographs were then taken of
samples 4, 6, 9, 10, 14, and 15 at magnifications of 100x, 200x, and 400x. These
photomicrographs (Figure 13 is an example) were used to characterize the phases present in the
slags (refer to the chapter on Results, section on Identification of Cogenetic Phases Through
Photomicrography). Four phases were identified in all the slag samples: (1) a silicon-melt phase,
(2) a quartz-like phase, (3) a copper phase, and (4) a copper-iron-sulfide phase (refer to
Appendix C for more examples of photomicrographs of the El Manchon slags).
Figure 13. A photomicrograph of sample 9B.
Electron Beam Microanalysis
Electron beam microanalysis was performed on samples 22, 23, 24, and 25 (see Table 1 in the
chapter on Methods, section on Sectioning and Photomicrography) to determine the elemental
compositions of the four phases identified. Electron beam microanalysis uses a finely focused
beam of electrons to probe a semi-spherical volume on the surface of a sample. The electron
beam causes the sample to emit characteristic x-radiation that can be used to identify the
elemental composition. A scanning electron microscope (SEM) images the surface of the sample
so that the electron beam can be focused on a particular point (N. Chatterjee, EAPS/MIT,
personal communication, August 2007). In this work electron beam microanalysis was used to
gather energy-dispersive spectra (EDS) which determine a qualitative analysis of the
composition present at a particular point. The electron probe from the Earth, Atmospheric, and
Planetary Science Department of the Massachusetts Institute of Technology was used. This
electron probe is a JEOL-JXA-733 Superprobe manufactured by JEOL Ltd.
Dr. Nilanjan Chatterjee of MIT's Department of Earth, Atmospheric, and Planetary Science and
Dr. Glenn Gaetani of the Department of Geology and Geophysics at the Woods Hole
Oceanographic Institute supervised my work with the electron probe.
The same 1" round mounts used for photomicrography were also used for electron beam
microanalysis. However, the 1" rounds had to be coated with a thin layer, approximately 20 nm
thick, of carbon in order to prevent charge accumulation during probing (Gaetani, personal
communication, August 2007). (Please see the chapter on Results, Electron Beam Microanalysis
and Appendix D for details of the SEM images and EDS spectra).
Ion Beam Microanalysis
Ion beam microanalysis was performed on samples 24 and 25 (see Table 1 in the chapter on
Methods, section on Sectioning and Photomicrography) to quantify the abundances of different
isotopes in the four phases identified. Ion beam microanalysis uses an ion beam to produce a
plasma on a roughly circular area tens of microns in diameter on the surface of a sample. The
plasma is then passed through an energy filter and induced into a mass spectrometer, which
separates and counts the isotopes (N. Shimizu, personal communication, October 2007). Ion
beam microanalysis uses an SEM to image the surface of the sample being analyzed, which
allows the user to target specific areas of the sample. I used the Cameca IMS 3f ion probe at the
Woods Hole Oceanographic Institute Northeast Nation Ion Microprobe Facility to carry out this
research.
This work was performed under the supervision of Dr. Nobumichi Shimizu of the Department of
Geology and Geophysics at WHOI. The same 1" round sample mounts used for electron beam
microanalysis and photomicrography were used for ion beam microanalysis. The carbon coating
that had been applied for electron beam microanalysis was polished off with 0.02 micron
colloidal silica on an automatic polisher for 10 minutes. Before the samples could be analyzed,
they had to be coated with a layer of gold about 20 nm thick.
The ion beam microprobe was calibrated to measure
238U, 232Th, 138Ba,
and 208Pb. 138Ba was set
as the measurement reference isotope because of its high concentration in the phases, yielding
8200 counts/sec, more than four times the count rate of the other isotopes. The primary beam
current was set to 3 nanoamps, which produces a probe beam of about 20 microns in diameter.
The silica-rich phase and the quartz-like phase were analyzed in both slag samples. The two
metal phases -- the copper and copper-iron-sulfide phases -- were indistinguishable from one
another, therefore the data obtained cannot be attributed to either metal. In addition, the prills
were typically smaller than the 20 micron beam size, so that the surrounding phases affected the
data for the metal phases. (Please refer to the chapter on Results, section on Demonstration of
Fractionation for details of the ion probe data).
PhysicalSeparationof Cogenetic Phases
The first step in separating the phases of the slag samples was to break the slags apart. For this
step, I used an electron pulse disaggregator (EPD). This is a rare machine that was custom-built
by the lab technicians at WHOI. It consists of a series of capacitors connected to two steel plates
that are connected to ground. The capacitors are set up in series in a bath of mineral oil to
prevent oxidation. The steel plates are contained in a Teflon housing filled with water. The two
steel plates are oriented at a 450 angle to each other with an adjustable gap between them.
The sample is placed in between the two plates. The ten capacitors are then charged and
subsequently discharged to the steel plates. Because the plates do not touch, but the sample
provides the physical connection between them, and the electric pulse forces its way through the
sample. This high voltage travels along the grain boundaries in the sample, because they are the
path of least resistance. This in turn causes the sample to expand differentially and to break
along the grain boundaries (B. Peucker-Ehrenbrink, personal communication, November 2007).
As the sample breaks apart, the pieces fall through the gap between the two plates and into a
reservoir. If the entire sample has fallen through the two plates, the water provides a sufficiently
conducting medium for the electricity to jump between the two plates. Dr. Bernhard PeuckerEhrenbrink of the Department of Marine Chemistry and Geochemistry at WHOI runs the EPD.
For this work, we set the EPD to a voltage of 30 V at 1 Hz. Fifty shots were fired per round at a
period of 8 shots per second. The gap was set to its smallest size: 1/8". Samples 24 and 25 were
completely disaggregated.
The disaggregated grains of Samples 24 and 25 were examined by petrographic reflected light
microscopy to determine into what categories the grains could be hand separated. Hand
separation is a straightforward but time consuming process. Using fine tweezers and some kind
of magnifying lens or microscope, I separated the grains based on visual distinctions between the
sorting categories into five portions: metallics, purities, impurities, grains-with-inclusions, and
fines. Metallics are categorized according to their unique, round morphologies that would be
observed only in metallic material. Purities are grains composed of the characteristic dark bulk
of the slags, consisting mostly of the silicon-melt phase. Impurities are grains other than
metallics or purities. They included grains of white to reddish color that may be protolith or
pieces of the smelting furnace lining. Grains-with-inclusions are grains that are composed of
more than a single phase. The fines category is composed of grains too small to be hand-sorted
with tweezers. (For more information on the hand-sorted portions refer to the chapter on
Results, section on Physical Separation of Cogenetic Phases).
The final separation method I used was magnetic separation. For this work, a Frantz
electromagnetic dry separator was used. The magnetic separator consists of an adjustable
electromagnet whose magnetic field influences an aluminum trough that divides into two
channels. The trough is set up on an incline so that the grains poured down the trough pass by
the magnet. The magnet pulls the magnetic grains into one channel of the trough. The nonmagnetic grains run down the other channel. I performed magnetic separations on Sample 24
37
under the supervision of Mr. Jerzy Blusztajn of the WHOI Plasma Mass Spectrometer Facility.
(Please refer to the chapter on Results, section on Physical Separation of Cogenetic Phases for a
full discussion the magnetic separation).
Results
Initial Slag Selection
Before the slag samples were modified in any way, they were photographed to document their
condition directly after excavation. I categorized the slags following Rachel Sharp's 2003
classification: rough (Figure 1) or smooth (Figure 2). (For a full catalogue of the slag samples
used in this work, refer to Appendix B).
0
I
2
3
4
5
6
7
6
9
10
Sample 6, bottom
Figure 14. An example of a smooth slag.
Sample 24, top
Figure 15. An example of a rough slag.
Sample Sectioning
The sectioning of the slag samples was carried out to facilitate mounting in 1" aluminum rounds.
I sectioned some of the larger samples twice (Table 2). Of the initial collection of 25 slags, only
10 of the samples were sectioned and mounted successfully. During the process, some of the
samples and mounts were damaged or were unsuitable for preparation with an automatic
polisher. Samples that could not be polished automatically were set aside because hand
polishing would have been too time consuming for the scope of this work. However, the
mounted samples still form a representative selection of smooth and rough slags from El
Manchon.
Table 2. Summary of sample sections.
Sample Number
Sharp's Categorization
Number of Sections
4
6
9
10
14
15
22
23
24
25
Smooth
Smooth
Rough
Rough
Rough
Rough
Rough
Rough
Rough
Rough
1
1
2
2
1
2
1
1
1
1
Identification of Cogenetic Phases Through Photomicrography
Using photomicrography, I identified four cogenetic phases in the El Manchon slags: a siliconmelt phase, a quartz-like phase, a copper phase, and a copper-iron-sulfide phase. These phases
were consistent in their morphologies between both the rough and smooth slag samples, but they
varied in their relative amounts within each sample. The smooth slag samples are composed
primarily of the silicon-melt phase; the other phases are present in small amounts. This internal
homogeneity gives rise to the even texture of the smooth slags. The rough slags exhibited a
much more heterogeneous internal composition, which causes the external rough texture of these
slags.
The first phase identified is the silicon-melt phase. This phase is the primary constituent of both
the rough and smooth slags, although it is present in smaller amounts in the rough slags. This
phase is composed of crystalline needle-like growths as well as amorphous zones (Figure 16).
The degree of crystallinity in the silicon-melt phase varies between slag samples as a function of
their different cooling rates. Slower-cooled examples of the silicon-melt phase exhibit a higher
degree of crystallinity, as the slow-cooling crystals had more time to grow (Figure 17). The
needle-like crystal growths are characteristic of iron silicates and were identified as fayalite
(2FeOoSiO 2) by Rachel Sharp (Sharp 2003).
b~iir,
~"~ii~
~~
--~
~ .i ~-;-r
--I -I-~~1
,4i-
Figure 16. A photomicrograph of the silicon-melt phase, showing amorphous and
crystalline zones.
Figure 17. A photomicrograph of dendritic growth in the silicon-melt phase. Dendritic
growth is characteristic of crystalline phases. This slag likely cooled sufficiently for
crystalline growth to begin, then solidified before crystal growth was complete.
The second phase identified is a quartz-like phase, so-called because of its resemblance to quartz
in section (Figure 18). The quartz-like phase is present in the rough slags in much greater
quantity than in the smooth slags, where it is present as only a minor constituent.
Figure 18. A photomicrograph of the quartz-like 1
in the voids in the quartz-like phase.
The other two phases identified are metallic phases: a copper phase and a copper-iron-sulfide
phase (metallurgical matte) (Figures 19 and 20). Both of these phases exist as round bodies,
referred to as prills, embedded in the silicon-melt phase. A matte is a phase formed during
metallurgical processing which contains copper, iron, and sulfur. The matte is a sulfide phase
and can be present as prills or in irregular shapes. The copper-iron-sulfide phase (matte) also
exists in irregularly shaped forms. Both of these phases exist only as minor components of the
slag samples. The copper phase has a rosy color unique to metallic copper. Many examples of
the copper phase also exhibited greenish internal surface corrosion characteristic of copper
43
carbonate or copper chloride corrosion products. The copper-iron-sulfide matte has a silvery
color, making it possible to distinguish the two metal phases visually.
rIgure iv. A pnotomicrograpn snowmg a copper
sulfide matte.
Sample 15A
200x
Copper-Iron-Sulfide Matte
Copper-Iron-Sulfide Prill
Copper Prill
Figure 20. A photomicrograph showing an atypically large example of a copper-ironsulfide matte. Copper and copper-iron-sulfide prills are also visible in the silica-melt
phase.
Compositional Analysis of Cogenetic Phases
I used electron beam microanalysis to characterize the qualitative element compositions of the
four cogenetic phases in samples 22, 23, 24, and 25. The elemental compositions of the four
phases determined by electron microanalysis were consistent between samples, which allowed
for elemental characterization of each phase. Some elements cannot be measured with the
electron microprobe; oxygen is an example of one such element. Table 3 indicates the number of
points analyzed in each slag sample, the phase analyzed, and the beam size, in microns.
Table 3. Summary of data points taken by electron beam microanalysis.
Probe Sample Phase Probed
Beam Size
Number
(microns)
22.1.1
22.2.1
22.2.2
22.2.3
22.2.4
22.2.5
22.3.1
22.3.2
22.4.1
22.5.1
23.1.1
23.1.2
23.1.3
23.2.1
23.2.2
23.2.3
24.1.1
24.1.2
24.1.3
24.1.4
24.2.1
24.3.1
25.1.1
25.1.2
25.1.3
25.2.1
25.2.2
Copper-Iron-Sulfide Phase
Silicon-melt Phase
Silicon-melt Phase
Silicon-melt Phase
Quartz-like Phase
Quartz-like Phase
Epoxy Bubble
Epoxy Bubble
Copper-Iron-Sulfide Phase
Copper-Iron-Sulfide Phase
Silicon-melt Phase
Silicon-melt Phase
Copper-Iron-Sulfide Phase
Copper-Iron-Sulfide Phase
Quartz-like Phase
Quartz-like Phase
Copper Phase
Silicon-melt Phase
Quartz-like Phase
Quartz-like Phase
Copper-Iron-Sulfide Phase
Copper-Iron-Sulfide Phase
Silicon-melt Phase
Copper Phase
Quartz-like Phase
Copper-Iron-Sulfide Phase
Copper Phase
1
20
10
30
10
30
10
20
1
1
10
20
1
1
10
30
1
10
10
30
1
1
10
1
10
1
1
The elemental composition of the silicon-melt phase consists primarily of silica with a
substantial presence of iron and smaller amounts of other elements (Figure 8).
Figure 21. The EDS spectrum for point 22.2.1. This is an example of a characteristic spectrum
for the silicon-melt phase.
The EDS spectra of the quartz-like phase showed the presence of only silica (Figure 22).
Presumably, oxygen is a primary constituent as well, but oxygen cannot be detected on EDS
spectra.
Figure 22. The EDS spectrum for point 23.2.2. This is an example of a characteristic spectrum
for the quartz-like phase.
During electron beam microanalysis, the backscattered electron images used to direct the
electron beam did not distinguish readily between the copper and copper-iron-sulfide phases.
However, the EDS spectra for the metal phases revealed two different compositions consistent
with my initial identification (Figures 23, 24, and 25).
rigure z. i ne vLil spectrum irom point z.z. 1 witn copper ana tron peaxs identitled. l his is a
characteristic spectrum for the copper-iron-sulfide phase. Readings of silicon probably come
from the silicon-melt phase surrounding the prill.
Figure 24. The EDS spectrum trom point 25.2.1 with iron peaks identified. This is a
characteristic spectrum for the copper-iron-sulfide phase. Readings of silicon probably come
from the silicon-melt phase surrounding the prill.
Figure 25. The EDS spectrum from point 25.1.2 with copper peaks identified. This is a
characteristic spectrum for the copper phase. Readings of silicon probably come from the
silicon-melt phase surrounding the prill.
Demonstration of Fractionation
Ion beam microanalysis was used to evaluate the degree of fractionation among the cogenetic
phases in Samples 24 and 25. The ion probe was calibrated to measure 238 U,
1 38
Ba. Stable
232 Th
232Th, 208Pb,
would behave chemically in the same manner as radioactive
230Th,
and
one of the
isotopes that I use to construct an isochron for the El Manchon slags. 226Ra is the isotope of
radium of interest in this work, but it is not measurable by ion beam microanalysis, because the
concentration of radium is too low. Therefore
138Ba
is measured in its place.
138Ba
is a stable
isotope that is not part of the uranium decay chain, but it is assumed to behave chemically like
radium.
208Pb
is not a daughter isotope of the uranium decay chain, but it behaves chemically in
the same manner as the isotopes of lead that are part of the uranium decay chain.
The backscattered electron images, which were acquired on the MIT electron microprobe, were
used to select the regions for analysis with the ion beam microprobe. These images showed
insufficient contrast between the metallic phases, therefore, the copper and copper-iron-sulfide
phases were classified as one category for ion beam microanalysis.
Table 4 records the fractionation of thorium and barium (i.e. radium) in the various phases of
Samples 24 and 25. The abundances of each element were normalized to the abundance of 38' Ba;
therefore the intensity of the
138
Ba readings is shown in the table.
Table 4. Summary of data points gather be ion beam microanalysis.
Probe Sample
232Th/13Ba
U/Th
138Ba/sPb
Number
Average
Intensity of
Ba (Kcps)
24.SiMelt. 1
24.SiMelt.2
24.SiMelt.average
24.Quartz.1
24.Metal.1
25.SiMelt.1
25.SiMelt.2
25.SiMelt.3
25.SiMelt.average
25.Quartz. 1
25.Quartz.2
25.Quartz.3
25.Quartz.average
25.Metal.1
25.Metal.2
25.Metal.average
0.00079
0.00095
0.00087
0.00479
0.00076
0.00103
0.00106
0.00116
0.00108
0.00123
0.00134
0.00142
0.00133
0.00092
0.00097
0.00095
2.5443
2.6842
2.6207
1.6117
2.6842
2.6408
2.6887
2.6897
2.6738
2.6098
2.4925
2.4014
2.4962
2.4565
2.4742
2.4656
29.9850
26.5534
28.1650
78.9889
37.6648
114.2857
106.9519
93.1966
104.0583
89.0472
100.6036
97.5610
95.4806
108.3424
102.3541
105.2632
290
7
260
250
180
240
Fractionation occurred in both samples. Sample 24 shows a greater degree of fractionation than
Sample 25. A larger degree of fractionation in a sample indicates that the data points from that
sample, shown on a plot of the isotopic ratios
226Ra/ 138Ba
VS 230 Th/138 Ba will exhibit a large
spread along the x-axis, leading to a better linear fit for the isochron. However, both samples
exhibit sufficient fractionation to be suitable for uranium series dating (Table 5).
Table 5. Fractionation in Samples 24 and 25.
23
Probe Sample
232Th/ 1Ba
U32 Th
13Ba/2Pb
Number
Average
Intensity of
Ba (Kcps)
24.SiMelt.average
24.Quartz.1
24.Metal.1
0.00087
0.00479
0.00076
2.6207
1.6117
2.6842
28.1650
78.9889
37.6648
290
7
260
25.SIMelt.average 0.00108
25.Quartz.average 0.00133
25.Metal.average 0.00095
2.6738
2.4962
2.4656
104.0583
95.4806
105.2632
250
180
240
Physical Separation of Cogenetic Phases
After the samples were disaggregated by electron pulse dissaggregation (see the chapter on
Methods, section on Physical Separation of Cogenetic Phases for a discussion of EPD), the
grains were hand-sorted. Because of the time intensity of hand-sorting, only Sample 24 was
hand-sorted completely. I used hand-sorting to divide the sample into five portions: metallics,
purities, impurities, grains-with-inclusions, and fines. Metallics are categorized according to
their unique, round morphologies that would be observed only in metallic material. Purities are
grains composed of the characteristic dark bulk of the slags, consisting mostly of the silicon-melt
phase (Figure 26). Impurities are grains other than metallics or purities (Figure 27). They
included grains of white to reddish color that may be protolith or pieces of the smelting furnace
lining. Grains-with-inclusions are grains that are composed of more than a single phase (Figure
28). The fines category is composed of grains too small to be hand-sorted with tweezers.
I"'
Figure 26. A portion of the purity separation of Sample 24.
-",'2
S,,;oANb
I
N
1
2~li
2
t
Figure 27. A portion of the impurities separation of Sample 24.
Figure 28. A portion of the grains-with-inclusions separation of Sample 24.
I sorted the fines by magnetic separation with a Frantz electromagnetic dry separator. This
enabled me to group the fines into three degrees of magnetism. The entire bulk of the fines was
separated with a current setting of 0.1 amps. The non-magnetic portion of this first separation
was labeled Vial 1. The remaining magnetic portion was then separated with a current setting of
0.02 amps. The non-magnetic portion of this separation was labeled Vial 2, and the magnetic
portion was labeled Vial 3. Using a hand-magnet, the contents of Vial 3 and the hand-sorted
metallics were separated into magnetic and non-magnetic portions.
Through this combination of techniques, I sorted Sample 24 into a total of nine separate portions
(Table 6). The isotopic ratios of each of these separations will constitute a data point that I will
use to construct the isochron for this slag sample. Therefore, the physical separations of the El
Manchon slags are sufficient to produce a multipoint isochron.
Table 6. Summart of the physical separations of grains in Sample 24.
Sample Number
V24-1
V24-2
V24-3
V24-4
V24-5
V24-6
V24-7
V24-8
V24-9
Composition
Vial 1 (non-magnetic fines)
Vial 2 (non-magnetic fines)
Vial 3, magnetic portion of fines
Metallics, non-magnetic portion
Metallics, magnetic portion
Inclusions
Vial 3, non-magnetic portion of fines
Powdered purities portion
Powdered impurities portion
Weight (grams)
2.962
5.399
36.051
0.250
0.247
1.450
1.057
2.411
2.717
Conclusion
Archaeological Utility of Uranium Series Dating
This thesis is the first attempt to apply uranium series dating to any kind of archaeological
material. In demonstrating that uranium series dating can be used to provide a date for the El
Manchon slags directly, this work also demonstrates that uranium series dating can be used with
any archaeological smelting or refining slag provided that it meets the four criteria required by
the analytical regime. Uranium series dating would offer a new method by which archaeologists
can determine the dates of sites with evidence of ore smelting or metal refining. Of course, the
use of this new tool is still in an experimental stage. Analysis of a variety of slags with the same
rigor that characterizes the research reported in this thesis will determine if uranium series dating
is universally applicable to slag samples.
Dating the El Manchon Slags
The primary goal of this thesis has been to develop a method for determining whether the
smelting technology at El Manchon was preHispanic or postHispanic in date. At El Manchon,
archaeological dating methods, such as radiocarbon (absolute method) or pottery typing (relative
method), proved inadequate to date the smelting technology directly. In addition, radiocarbon
dates determined for other areas of the site yielded an age range that spans both preHispanic and
postHispanic eras (ca. 750 to 1850 CE). If the geological technique known as uranium series
dating can be adapted in order to provide dates for the El Manchon slags, it will be possible to
date directly the smelting technology that was performed at El Manchon.
The overall goal of uranium series dating is to produce an isochron experimentally. An isochron
is a plot of the normalized activities of two measurable radioactive isotopes in the cogenetic
phases of a slag sample that takes the form of a straight line (Figure 29).
Theoretical Isochron
3
5 2.5
2 Equiline
Experimentally derived
isochron
* Experimental data
points
"- 1.5
S1
S0.5
0
1
2
3
((230Th)/BaE-3) {(dpm/g)/(ug/g))
Figure 29. An example of theoretical data points in dark blue, with a constructed isochron line in
dark red. By comparing the slope of the isochron with the slope of the equiline, a date can be
determined for the sample.
By comparing the isochron to the equiline, the age of the slag sample can be determined. The
equiline is the plot of the normalized activities of two measurable radioactive isotopes in the
cogenetic phases of a material that has aged enough to achieve secular equilibrium between the
two isotopes. Because the rate at which isotopes decay to secular equilibrium is well understood,
a comparison between the isochron and the equiline allows an age to be calculated for a sample.
This thesis evaluated four criteria to determine whether or not uranium series dating through the
experimental establishment of an isochron would be appropriate for dating the El Manchon
slags: (1) measurable presence of appropriate isotopes in the slags; (2) cogenetic phases within
the slags; (3) isotopic fractionation between cogenetic phases; and (4) the physical separatability
of the cogenetic phases. The work of Dr. Kenneth Sims and Dr. Christian Miller at the Woods
Hole Oceanographic Institute demonstrated that adequate, measurable abundances of the
isotopes
2 30
Th and
Ra, the isotopes of interest for this study, are present in the El Manchon
22 6
slags.
My thesis research went further to demonstrate that the slags contain four distinct cogenetic
phases: a silicon-melt phase, a quartz-like phase, a copper phase, and a copper-iron-sulfide
phase. Measurements of the 226Ra/ 138 Ba VS 230Th/ 138Ba normalized activities within these phases
would yield an isochron with at least four points, sufficient to determine the date of the slags. I
measured the extent of isotopic fractionation within each of the four phases through ion beam
microanalysis. Finally, I separated slag Sample 24 experimentally into nine portions that
represent the four cogenetic phases, demonstrating that the El Manchon slags can be physically
separated successfully.
My research demonstrates that the El Manchon slags meet the four criteria established by
geochemistry and that they would be good candidates for uranium series dating. The age
determined for a piece of El Manchon slag provides the date when the slag solidified, thereby
effectively dating the smelting technology directly. While both rough and smooth slags
excavated at El Manchon met all four criteria, the rough samples are better suited to uranium
series dating. The smooth slags are almost homogenously composed of one phase, whereas the
rough slags contain greater quantities of the four phases. The presence of ample amounts of each
phase facilitates the analysis.
Further Research
I have begun the next step in my research, which involves the dissolution of El Manchon slag
Sample 24 in preparation for the column chemistry necessary to measure quantitative isotope
abundances in each phase separation. The column chemistry serves to purify the elements of
interest, in this case uranium, radium, thorium, and lead. Once the elements have been purified,
mass spectrometry will quantify the abundances of the isotopes of each of the elements. This
work is being carried out at WHOI under the supervision of Dr. Kenneth Sims, Department of
Geology and Geophysics. The calculation of a date for El Manchon slag Sample 24 will
demonstrate the pre- or postHispanic nature of the smelting carried out at El Manchon.
My results will offer archaeologists a new dating method central to the interpretation of
extractive metallurgical technologies of pre-industrial and prehistoric societies.
Appendix A: Mathematical Derivation of Isochron Slope
Let us consider a decay chain:
230Th > 226Ra
decays exponentially into its daughter isotope 226Ra. For the time frame of the isochrons
used for the El Manchon slags, 230Th is not significantly replenished, therefore 23 0Th decay is as
follows,
23 0Th
(1)
-dN230Th = k230Th N230Th
226Ra
decays exponentially as well, but is being replenished by 23 0Th,
dN.26Ra
(2)
= k230Th N230Th - X226Ra N226Ra
The amount of 23 0Th through time,
e-,230Th t
N230Th = N230Th
(3)
Substituting (3) into (2),
e-%230Th t
dN926Ra = k230Th No230Th
k226Ra N226Ra
(4)
dt
Setting (4 = 0),
dN226Ra - k230Th N0 230Th
e - 2 30 T h t+
226Ra N226Ra = 0
(5)
dt
Solving (5) for N226Ra
N0 230Th (e30Th t _ e-2 2 6 Ra t) + N0 226Ra e- 22 6 Ra t
N226Ra = k230Th
X226Ra -
(6)
230Th
Assuming there is no initial 2 26 Ra (N 0 226Ra = 0), (6) becomes:
N226Ra =
N 0 230Th (e-'230Th t _ e-2
230Th
k226Ra -
230Th
26
Ra t)
(7)
Assume that
X226Ra - k230Th = X226Ra,
and
e-230Th t
1
because the decay constant of 230Th is sufficiently low.
This makes (7)
N226Ra= k230Th N0 23 0Th(1 - e -k 22 6R a t)
(8)
X226Ra
Multiplying (7) by
X226Ra
X226Ra vN226Ra= k230Th N
2 3 0Th
(1 - e -
226
Ra t)
(9)
Substituting in activities,
A226Ra = A230Th (1 - e-
22 6
From (3), for initially present
A226Ra= A
0
22 6Ra
e-k
22 6
(10)
Ra t)
22 6
Ra
(11)
Rat
Now let us assume, there is decay of initially present 226Ra as well as decay of 226Ra produced by
230Th,
e - X2 2 6 Ra t)
(12)
Normalizing (12) by A138Ba,
A226Ra= A 226Ra ek22 6 Rat + A30Th (1 - e - 226Ra t)
(13)
A226Ra = A 0 226Ra e -
A128Ba
A128Ba
2 26
Rat
+
A
2 3 0Th
(1-
A128Ba
In a linear graph of (A226Ra/A138Ba) against (A230Th/A138Ba), an isochron, the slope of the line will
be:
(1 - e
2 26
Rat)
Because X226Ra is known, an age can be calculated from the slope of an isochron.
(14)
Appendix B: Initial Slag Sample Collection
All photographs were taken with a CoolPix 5400 digital camera set to macro mode, with 5M
pixel proportions and high quality. The photographs had a background of ground glass over
white paper.
Table 7. Summary of the slag samples used in this work.
Sample Type
Dimensions (cm)
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
5.5 x 4 x 0.25
6 x 4x 3
4x4x3
2.5 x 3 x 0.7
2.5 x 3 x 2.5
4.5 x 6 x 0.5
3 x 4 x 0.5
4 x 4 x 0.4
2 x 3.5 x 2
3 x 5 x 2.5
2.5 x 3 x 2.5
3.5 x 5.5 x 0.5
4 x 6.5 x 0.5
2.5 x 5.25 x 3
4.5 x 4.5 x 1
4 x 3.5 x 2.5
1.5 x 3 x 1.5
3.5 x 7 x 2.5
6.5 x 7 x 0.5
3 x 3 x 2.5
16 x 18 x 1
8x 7 x 2
7x 9x2
6 x 8x 2
7x9x2
Smooth
Rough
Rough
Smooth
Smooth
Smooth
Smooth
Smooth
Rough
Rough
Rough
Smooth
Smooth
Rough
Rough
Rough
Smooth
Rough
Smooth
Rough
Smooth
Rough
Rough
Rough
Rough
Shown Here
X
X
X
X
X
X
X
X
X
11M
IIIM
11111111MIM
-
Figure 30. Sample 4, top view
o
3
Sample 31. Sample 6, bottom view.
4
s
6
10
Figure 32. Sample 9, top view.
Figure 33. Sample 10, side C.
M
S4
IL
a
5
7
4
5
10
1
L
Figure 34. Sample 14, side C.
O
r
a
Figure 35. Sample 15, bottom view.
S
9
10O
Figure 36. Sample 21, top view.
Figure 37. Sample 24, top view.
Figure 38. Sample 25, top view.
Appendix C: Slag Sample Photomicrographs
This appendix provides a selection of photomicrographs from the El Manchon slags used in this
research.
Figure 39. A photomicrograph of a smooth slag. The main phase present is the siliconmelt phase.
Figure 40. A photomicrograph of a smooth slag. The main phase present is the siliconmelt phase.
Figure 41. Photomicrograph of a rough slag. The cracks of the quartzfilled with silicon-melt phase.
Figure 42. Photomicrograph of a rough slag. The quartz-like phase and the silicon-melt
phase are present.
Figure 43. Photomicrograph of a rough
present.
Figure 44. Photomicrograph of a rough slag. Dendritic growth is visible in the siliconmelt phase.
Figure 45. Photomicrograph of a rough slag. This sample has a predominant copper
prill and surrounding copper-iron-sulfide matte.
4x,
.
F
/
'
°4
Figure 46. Photomicrograph of a rough slag. The silicon-melt phase and quartz-like
phase are prominent.
0
*4
o
7
1B
<.
;
-.4
r
.
-lOB
A
Figure 47. A photomicrograph of a rough slag. A copper-iron-sulfide prill is visible in
the lower right.
Figure 48. A photomicrograph of a rough slag. The silicon-melt phase is prominent.
Figure 49. A photomicrograph of a rough slag, showing the quartz-like phase, siliconmelt phase, and copper-iron-sulfide phase.
I'; .
4'
9
"
7
Figure 50. Photomicrograph of a rough slag. An usually large copper-iron-sulfide
phase is visible.
~I
Appendix D: Electron Beam Microprobe Data
Table 8. Summary of data points analyzed by electron beam microprobe.
Probe Sample Phase Analyzed Beam Size
Number
(micr ns) I
22.1.1
22.2.1
22.2.2
22.2.3
22.2.4
22.2.5
22.3.1
22.3.2
22.4.1
22.5.1
23.1.1
23.1.2
23.1.3
23.2.1
23.2.2
23.2.3
24.1.1
24.1.2
24.1.3
24.1.4
24.2.1
24.3.1
25.1.1
25.1.2
25.1.3
25.2.1
25.2.2
Metal Phase
Silicon-melt Phase
Silicon-melt Phase
Silicon-melt Phase
Quartz-like Phase
Quartz-like Phase
Epoxy Bubble
Epoxy Bubble
Metal Phase
Metal Phase
Silicon-melt Phase
Silicon-melt Phase
Metal Phase
Metal Phase
Quartz-like Phase
Quartz-like Phase
Metal Phase
Silicon-melt Phase
Quartz-like Phase
Quartz-like Phase
Metal Phase
Metal Phase
Silicon-melt Phase
Metal Phase
Quartz-like Phase
Metal Phase
Metal Phase
1
20
10
30
10
30
10
20
1
1
10
20
1
1
10
30
1
10
10
30
1
1
10
1
10
1
1
30pm
BSE spot22pl
Figure 51. Photomicrograph of Sample 22, frame 1. Copper-iron-sulfide phase
identified by red point.
Figure 52. EDS spectrum for Point 22.1.1. Sulfur, copper, and iron are the major constituents.
Figure 53. EDS spectrum of Point 22.1.1 with characteristic Cu peaks identified.
Figure 54. EDS spectrum of Point 22.1.1 with characteristic Fe peaks identified.
~
1;5
I~I ~r6~
;
.
i
I
-5
~X
1lli~
-~iff~~
iii~E :; ;
r
i
Point22.2.1
. Point 22.2.2
Point 22.2.3
300pm
BSE spot22p2
Figure 55. Photomicrograph of Sample 22, frame 2. Silicon-melt phase
identified by black point; quartz-like phase identified by blue point.
Figure 56. EDS spectrum of Point 22.2.1. Silicon and iron are the main constituents.
Figure 57. EDS spectrum for Point 22.2.1 with characteristic Fe peaks identified.
Figure 58. EDS spectrum for Point 22.2.2. Silicon and iron are the main constituents.
Figure 59. EDS spectrum for Point 22.2.2 with characteristic Fe peaks identified.
Figure 60. EDS spectrum for point 22.2.3. The main constituents are silica and iron.
igure oi.
Nspectrum tor point zz.z. wltn cnaractenstic iron peaks identitled.
Figure 62. EDS spectrum for point 22.2.4. The primary constituent is silica.
Figure 63. EDS spectra for point 22.2.5. The primary constituent is silica.
300pm
BSE spot22p3
Figure 64. Photomicrograph of Sample 22, frame three. An epoxy zone is identified by the blue
point.
Figure 65. EDS spectrum for 22.3.1. Chlorine and silica are major constituents.
Figure 66. EDS spectrum for point 22.3.1 with characteristic chlorine peaks identified.
Figure 67. EDS spectrum of point 22.3.2. The major constituents are chlorine and silica.
Figure 68. EDS spectrum of point 22.3.2 with characteristic chlorine peaks.
300pm
BSE spot22p4
Figure 69. Photomicrograph of Sample 22, frame 4. Copper-iron-sulfide phase identified by blue
point.
Figure 70. EDS spectrum for point 22.2.4. Copper, iron, and sulfur are the main constituents.
Figure 71. EDS spectrum for point 22.2.4 with characteristic copper peaks identified.
300pm
BSE spot22p5
Figure 72. A photomicrograph of Sample 22, frame 5. Copper-iron-sulfide phase identified by
blue point.
Figure 73. EDS spectrum of point 22.5.1. The main constituents are iron, silica, and
copper.
Figure 74. EDS spectrum of point 22.5.1 with characteristic iron peaks identified.
Figure 75. EDS spectrum for point 22.5.1 with characteristic copper peaks identified.
300pm
BSE spot23pl
Figure 76. Photomicrograph of Sample 23, frame 1. Silica-melt phase identified by black dot.
Copper-iron-sulfide phase identified by red dot.
Figure 77. EDS spectrum of point 23.1.1. Iron and silica are the main constituents.
Figure 78. EDS spectrum of point 23.1.2. The main constituents are iron and silica.
Figure 79. EDS spectrum for point 23.1.2 with characteristic iron peaks identified.
Figure 80. EDS spectrum for point 23.1.3. The main constituents are iron, copper, and sulfur.
Figure 81. EDS spectrum for point 23.1.3 with characteristic iron peaks identified.
Figure 82. EDS spectrum of point 23.1.3 with characteristic copper peaks identified.
Point 23.
Point 23.2.3
300pm
BSE spot23p2
Figure 83. Photomicrograph of Sample 23, frame 2. Copper-iron-sulfide phase identified by black
point. Quartz-like phase identified by blue point.
Figure 84. EDS spectrum of point 23.2.1. The major constituents are iron, sulfur, and copper.
Figure 85. EDS spectrum for point 23.2.1 with characteristic iron peaks identified.
Figure 86. EDS spectrum for point 23.2.1 with characteristic copper peaks identified.
Figure 87. EDS spectrum of point 23.2.2. The major constituent is silica.
Figure 88. EDS spectrum of point 23.2.3. The major constituent is silica.
300pm
BSE spot24pl
Figure 89. Photomicrograph of Sample 24, frame 1. Copper phase identified by black point.
Silica-melt phase identified by purple point. Quartz-like phase identified by blue point.
Figure 90. EDS spectrum of point 24.1.1. The main constituent is copper.
Figure 91. EDS spectrum for point 24.1.1 with characteristic copper peaks identified.
Figure 92. EDS spectrum of point 24.1.2. The major constituents are iron and silica.
0
M
M
Figure 93. EDS spectrum for point 24.1.2 with characteristic iron peaks identified.
Figure 94. EDS spectrum for point 24.1.3. The major constituent is silica.
Figure 95. EDS spectrum of point 24.1.4. The major constituent is silica.
300pm
BSE spot24p2
Figure 96. Photomicrograph of Sample 24, frame 2. Copper-iron-sulfide phase identified by blue
point.
Figure 97. EDS spectrum of point 24.2.1. The main constituents are iron, copper, and sulfur.
Figure 98. EDS spectrum for point 24.2.1 with characteristic iron peaks identified.
Figure 99. EDS spectrum for point 24.2.1 with characteristic copper peaks identified.
BSE spot24p3
Figure 100. Photomicrograph of Sample 24, frame 3. Copper-iron-sulfide phase identified by
blue point.
Figure 101. EDS spectrum for point 24.3.1. The major constituents are iron, copper, and sulfur.
Figure 102. EDS spectrum for point 24.3.1 with characteristic iron peaks identified.
100
Figure 103. EDS spectrum for point 24.3.1 with characteristic copper peaks identified.
300pm
BSE spot25pl
Figure 104. Photomicrograph of Sample 25, frame 1. Silica-melt phase identified by the black
point. Copper phase identified by the purple point. Quartz-like phase identified by the blue
point.
101
Figure 105. EDS spectrum of point 25.1.1. The major constituents are silica and iron.
Figure 106. EDS spectrum for point 25.1.1 with characteristic iron peaks identified.
102
Figure 107. EDS spectrum for point 25.1.2. The main constituent is copper.
Figure 108. EDS spectrum of 25.1.2 with characteristic copper peaks identified.
103
Figure 109. EDS spectrum for point 25.1.3. The main constituent is silica.
300pm
BSE spot25p2
Figure 110. Photomicrograph of Sample 25, frame 2. Copper-iron-sulfide phase identified by
purple point. Copper phase identified by blue point.
104
Figure 111. EDS spectrum for point 25.2.1. The major constituents are copper, iron, and sulfur.
Figure 112. EDS spectrum for point 25.2.1 with characteristic iron peaks identified.
105
Figure 113. EDS spectrum for point 25.2.1 with characteristic copper peaks identified
Figure 114. EDS spectrum for point 25.2.2. The main constituent is copper.
106
Figure 115. EDS spectrum for point 25.2.2 with characteristic copper peaks identified.
107
Acknowledgements
I would like to thank first and foremost my thesis advisor, Professor Dorothy Hosler
(DMSE/MIT) for giving me the opportunity to work on this project and arranging for me to carry
out research at the Woods Hole Oceanographic Institute (WHOI). The idea of using the uranium
series for dating archaeological slags was suggested to Prof. Hosler by Dr. Andrew Macfarlane
(Department of Earth Science/Florida International University).
I would also like to thank Dr. Kenneth Sims (Geology & Geophysics/WHOI), my direct
supervisor at WHOI.
I would like to thank Prof. Heather Lechtman (DMSE/MIT) and Dr. Sidney Carter
(CMRAE/MIT) for editorial suggestions during the writing process.
I would also like to thank
Dr. Nilanjan Chatterjee (EAPS/MIT)
Dr. Glenn Gaetani (Geol. & Geophys./WHOI)
Dr. Nobumichi Shimizu (Geol. & Geophys./WHOI)
Dr. Bernhard Peucker-Ehrenbrink (Geol. & Geophys./WHOI)
Mr. Jerzy Blusztajn (Plasma Mass Spectrometer Facility/WHOI)
Ms. Jennifer Meanwell (DMSE-CMRAE/MIT)
and
Mr. Christian Miller (Geol. & Geophys./WHOI)
Ms. Hannah Reitzel (DMSE/MIT)
Mr. Christopher Waters (Geol. & Geophys./WHOI)
Ms. Evelyn Mervine (Geol. & Geophys./WHOI)
Ms. Malima Wolf (Mechanical Engineering/MIT)
I would like the Undergraduate Research Opportunities Program for financial support during the
summer of 2007.
108
References Cited
Bachmann, Hans-Gert
1982
The Identification of Slags from ArchaeologicalSites. The Institute of
Archaeology, Occasional Publication No. 6, London, England.
Cooper, Kari M., Steven J. Goldstein, Kenneth W.W. Sims, and Michael T. Murrell
2003 Uranium-series chronology of Gorda Ridge volcanism: new evidence from the
1996 erruption. Earth and Planetary Science Letters 206, No. 3: 459-475.
Faure, Gunter and Teresa M. Mensing
2005 Isotopes PrinciplesandApplications. 3 rd Ed. John Wiley & Sons, Inc., Hoboken,
New Jersey.
Hedman, Matthew
2007 The Age ofEverything: How Science Explores the Past. The University of
Chicago Press, Chicago, IL.
Hosler, Dorothy
2004 Nuevos datos sobre la producci6n del metal en el occidente en la 6poca
prehispinica. Bienes Regionales del antiguo Occidente de Mixico: Producci6ny
Intercambio. Colegio de Michoacin, pp. 335-354.
2003a "Metal Production," Chapter 21 in The PostclassicMesoamerican World.
University of Utah Press: 159-171.
2003b Nuevos hallazgos sobre la metalurgia antigua de Guerrero. El Pasado
Arqueol6gico de Guerrero. Instituto Nacional de Antropologid y Historia.
Mexico D.F., pp. 225-241.
1994 The Sounds and Colors ofPower: The Sacred MetallurgicalTechnology of
Ancient West Mexico. MIT Press, Cambridge, MA.
1988a Ancient West Mexican Metallurgy: A Technological Chronology. Journalof
FieldArchaeology 15, No. 2: 91-217.
1988b Ancient West Mexican Metallurgy: South and Central American Origins and
West Mexican Transformations. American Anthropologist 90, No. 4: 32-855.
Ivanovich, M. and R.S. Harmon, eds.
1982 Uranium Series Disequilibrium: Applications to EnvironmentalProblems.
Clarendon Press, Oxford, UK.
Sharp, Rachel
2001 Analysis of Copper Slags from the ArchaeologicalSite ofEl Manchon, Guerrero,
Mexico. Bachelor of Science Thesis, Massachusetts Institute of Technology.
109